Using A Mixing Desk To Combat Latency through DAW

[TC1]

Hi. I'm new here and have had a look to see if this has been brought up recently, so apologies in advance if it is has, and i've missed it. I know it is a much discussed topic (generally.)

I'm looking for simple advice on how to I would use a mixing desk to comabt latency in a DAW based recording solution. My basic setup would go like this, saffire pro 40, octo-pre (via optical into saffire into DAW).

I have access to a soundcraft ghost analogue mixer and am looking for (very simple) instructions on how to use the desk to minimise latency/ achieve 'zero' latency. I think this is done by using a y lead at the insert points in order to split the signal from the preamps into the desk and into the DAW at the same time, but am not sure if this is correct? I want to be able to monitor through the DAW (use software monitoring), i.e. hear the signal as it is, going onto disk. I could simply bypass the software moniroting in the DAW, but am nervous about doing this as I don't like to hear it going in only and not be able to quality check the output from the DAW for usunsual glitches for example. Is it possible to do this by using the described method, intergrating an analogue desk into the heart of the systme in order to monitor? I don't want to use the desk as part of the recording chain. I want to use the pre-amps on the octo-pre and saffire, just monitor through the desk to get rid of latency. I would need an idiots guide on how to do this, so thanks in advcne for any help, which would be greatly appereciated.

[theblue1]

There are, of course, a number of places where digital processing latency raises its ugly head.

Generally, the most obvious is the kind of monitoring latency you address here. (One of the other problems is that of tracking misalignment due to unadjusted-for latency in the tracking process, apparently typically from an audio interface device driver mis-reporting its own latency. One tests for it using a ping-loopback test. It's a somewhat complex issue beyond the scope of this post.)

You can use an analog mixer for latency-free input monitoring, as well as an intermediary form of volume control between your converter's output and powered monitors (also eliminating the need to control output volume from a digital-side volume control in your computer and the possible loss-of-audio-resolution issues when using a digital volume control at low volumes).

You should be able to find diagrams and info on such a set up here:

How to Hookup a Mixer

[shanabit]

Am I missing something here? Your Saphire has zero monitoring on it

"Saffire PRO 40 Control Zero-latency DSP Mixer/Router
Routing flexibility and intuitive one-click set-up solutions; Saffire PRO 40 Control provides an 18 x 16 DSP mixer with excellent output routing and monitoring capabilities.

[theblue1]

Sadly, a number of gear makers (not sure Focusrite is one of them but I strongly suspect it is) use the term "zero latency monitoring" extremely loosely.

There is no such thing as zero latency digital monitoring, in my book. The AD and DA process used in the near-zero latency monitoring in most contemporary digital audio interfaces take roughly a millisecond apiece, give or take, meaning that the lowest monitoring latency for such onboard "zero latency" monitoring is actually a moderately significant 2 ms.

And 2 ms is not zero.

Where I come from.

[But it is apparently where my interface came from. MOTU (from whom I won't be buying again, thank you) apparently don't care any more about marketing speech accuracy than other outfits. My 828mkII is marketed as having "zero latency" onboard monitoring. Which is, by my definition, a lie. But it's one that is widely told by a number of makers; so, when you see "zero latency," remember -- it doesn't mean shit.]


Now, 2 ms is certainly better than the 5 to 15 ms latency many USB and FW interface users experience (and even some PCI card users) and may well be close enough for some musicians on some instruments. For instance, I don't notice the latency in my MOTU's onboard cue mix when I'm cutting my own vocals or playing an acoustic guitar because I can hear the sound of my voice or the guitar and the tiny lag is, in a sense, 'covered up' by the actual sound of my voice or guitar in the room. OTOH, if I try playing my solid body electric d.i. into the MOTU [which I don't even bother with anymore, though the inst input wasn't too bad -- when it worked, but a couple years after I bought the MOTU, it developed a nasty 'ticking' noise in the mic/inst preamps; POS], the tiny lag is decidedly noticeable. I can play around it, it is a tiny lag, but it is definitely noticeable as a feel thing -- it's as though the guitar was 'sluggish.'

[TC1]

Hi. Thanks to you both for taking the time to read the thread and get back to me.

Those diagrams are extremly helpful, thank you. Yea, re the zero latency claim, I don't really believe it exists either (i've yet to buy the two units). I have a feeling that the zero latency claim wil just be really a preset in the saffire control to bypass the software, which i find undesirable. I've read that the 'round trip' latency on the saffire is around 13ms, by the time the octo-pre goes through an A/D conversion into the saffire, the round trip latency wil be fractionally more. This of course may be wrong and the latency on the saffire may be perfectly acceptable. I'd just like a full proof way of making it absolutely acceptable using a desk if possible as it is an area I haven't explored before. I have the original (two channel) Saffire, and whilst I think the pre-amps were excellent for the price, it is listed as low latency and, basically monitoring through software is impossible with this otherwhise excellent little unit.

You have mentioned recording files misalining with each other. I have experienced this using my original two channel Saffire, which has caused some embarrassment at playback. Is there any way to get around this little problem? It actaully seemed to rear it's head out of nowhere after it had been working fine in previous uses. (Thanks again all this is hugely appreciated)

[theblue1]

Some interfaces do have a built-in analog source/playback mixer -- which would provide true zero latency monitoring -- but most have a built in digital mixer that takes the playback from the computer and mixes it with the source -- after the source has gone through AD -- and then that all goes back out through DA. Which I would call near-zero latency -- but which many manufacturers call "zero latency," even though it's probably a couple ms.

But where you get the really noticeable latency is when your source signal makes the full round trip from the interface over a bus (whether it's USB, FW, or even the slightly lower latency PCI bus that add-in cards typically use).


With regard to the tracking misalignment issue -- you can test to see if you have such problems with a so-called "ping loopback test." Essentially you take a wave with an easily identifiable starting place. (I use a 1kHz sine tone with arelatively low amplitude, say around -18 dB [a nice round figure... you basically just don't want to get blasted if you accidentally listen to the test tone].)

Output your test tone playback in your DAW, with the analog output routed to an analog input and record that into another track. (Turn off source monitoring and turn down the CR volume to be on the safe side or you can end up with painful, damaging feedback if you're careless.)

In a perfect world, the orginal test tone and the new copy will line up to the sample on the timeline.

But in all likelihood (based on people I've got to test their rigs), unless you've otherwise compensated, you'll likely see a timeline misalignment of the two signals, with the copy most likely coming 'after' the original on the timeline.

If your DAW doesn't have a way of compensating for such a misalignment, you'll have to manually slide or nudge your new overdubs by the amount of difference between the original and copy.

Happily, many contemporary DAWs have such a capablity. AIUI, Reaper, Cubase and Sonar, and Tracktion, among others, have an automated time alignment calibration utility. (Sometimes called hardware/tracking alignment adjust and such. Basically it's a way of automating a ping loopback and setting an offset based on the differential.)

Now, a problem that sometimes arises is that a given device will have a tracking misalignment that varies from session to session. (I had a USB mic that was sometimes ~25 ms out and sometimes ~40 ms out.)

Another problem can arise if the interface's driver is so wack that it overcompensates for latency and a given sound is too 'early' on the timeline. In some DAWs, there is no way short of manually moving each new overdub by the difference, if the overdub is 'early' instead of 'late.'


In my experience, many folks in the past suffered from such problems but seemed completely oblivious to them.

In fact, I had huge problems initially even explaining this issue to some very experienced, clued in people. I remember describing the loopback test elsewhere and having some folks -- seasoned pros, I'm afraid -- saying things like, "I never record the analog outputs of my DAW into the analog inputs of it -- so why would I care if they line up?"

Stuff like that just made my jaw drop... How on earth do those people get through an average day of dealing with technology? I guess it shows you can be a real slow thinker and still be a pro in this industry.


Me... I've long wondered if precisely this very issue might not be at the heart of what many people don't like about recording digitally.

My own MOTU 828mkII (using WDM/KS drivers that work best for me with it; ASIO drivers glitch out on my system no matter how I set them up) has an uncorrected offset of nearly 8 ms...

While mostly one probably works from a central drum track in larger projects, and so that would provide something of a rhythmic center, my favorite example of what can go really, really wrong by not correcting such misalignment would be something like this:

drums go down first; bass player plays to the time he hears over his monitors; on my rig, that gets laid down 8 ms behind what he's hearing; the keyboardist then comes in and listens to the bass player, not the drums, and he's now 16 ms behind the drums; johnny guitar comes in and listens to the keyboardist... his part is now 24 ms behind the drums... You get the drift.

It's a rather extreme example, but it highlights the issue rather nicely I think.

[Lawrence]

ASIO DM uses the converters.



Some mfgs control panels may allow a direct analog patch to output which would take those 1-2ms out of the equation.

[theblue1]

[First off -- great illustration! You can bet it will pop up again and again around here. I can just about guarantee it.]


But, yeah, right, I wanted to avoid that whole speed of electricity thing. There are a lot of reasons that digital processes take time and practical reasons why we use buffering transmission protocols for many of the digital signals we pass from one system to another.


But back to the zero latency monitoring claims made by some manufacturers:

There are a number of converter makers who use digital cue mixes (like my very own MOTU 828mkII [not buying MOTU again; this is not an endorsement]) yet who claim that that produces what they call "Zero Latency Monitoring."

I've heard the term "direct monitoring" used even in software, so I'm afraid that that term can be considered a bit suspect, as well.

But if a given unit claims true all analog zero latency or direct monitoring, then I think it's pretty much good to go.


The problem is when the suits up in marketing start monkeying with the truth.

[shanabit]

my MOTU 828 MKii allows your to monitor any of the inputs directly out the main outs or headphone outs WITHOUT connecting it to a computer so its the SAME as an analog console in that respect. No diff than using an old ADAT deck and a console while monitoring it.

There IS NO audible delay when I use it this way. I can route a hardware effects box from it as well during tracking since it has a send on it.

[theblue1]

I'm not actually familiar with the earlier MOTU 828.*

My 828mkII definitely uses a digital cue mix. Hence the name CueMix DSP. (Of course, that all functions whether or not it's connected to a computer. It's a little built in digital mixer.)

And on this page, they actually address the issue more or less appropriately, by use of a nicely vague qualifier: "virtually no monitoring latency" -- MOTU.com - 828mkII Overview [bold added]

Virtually... like in... almost... not much... near...

To be sure... it's not a lot. Probably 2 ms or under. Certainly nothing like the ~8 ms monitoring latency I get going to the computer and back -- which is a painfully noticeable dealbreaker.


Unfortunately... the liars in suits got hold of the marketing copy for the 828mk3 -- and look what happened to the copy:

"Record, monitor, route and process all of these live inputs using the professional on-board CueMix FX digital mixer – with no latency and no processor strain on your computer." MOTU.com - 828mk3 Overview

[bold added to highlight contradictory claims]


Like I said... I won't be buying MOTU again. In any foreseeable future.

__________________

* this bit from a review of the orginal MOTU 828 from a 2001 Electronic Musician seems to contradict what you say about that device's cue mixing:

The 828's control panel also provides access to CueMix Plus, which lets you mix the signal going into any analog input with the main outputs in real time. The monitoring latency of CueMix Plus is lower than the original CueMix that came with earlier MOTU audio interfaces. A dedicated monitor-level knob balances the selected live input with the main outputs, and the main Volume knob controls the summed signal.

But... elsewhere, the same review says:

Zero-latency monitoring of any analog input or pair.

MOTU 828 (MAC/WIN)

Does the reviewer (Rob Shrock) mean that there is a cue mix separate from CueMix Plus? And is that limited to monitoring just one analog input pair at a time? I seriously doubt that there is a full analog summing matrix in addition to the digital one... that seems highly unlikely.

[Gordon]

next silly question - it's never going to get better is it? I have a suspicion that converter "speed" isn't even a priority with manufacturers. There will always be latency if you ain't using an analog mixer.

Now "noticeable" is the operative word. I just picked up a MOTU 828 MKIII last night and I noticed nothing playing my bass through the quemix. At a 64-sample Buffer my round-trip latency clocks at 5.5 ms which I can clearly feel. it's okay, except when i slap the lag is obvious. I know MOTU is know for having good round-trip latency with the only thing better in the Firewire realm being RME which can go down to s 32-sample buffer.

It's not ever going to get better is it?

edit - you're right, the MOTU blows. Comparing side-by-side with my 10-year old Mixtreme the MOTU's internal mixers latency sucks. It can definitely feel it when I shred/slap on my bass. I'm returning this piece of garbage tomorrow.

[Gordon]

Originally Posted by shanabit

my MOTU 828 MKii allows your to monitor any of the inputs directly out the main outs or headphone outs WITHOUT connecting it to a computer so its the SAME as an analog console in that respect. No diff than using an old ADAT deck and a console while monitoring it.

There IS NO audible delay when I use it this way. I can route a hardware effects box from it as well during tracking since it has a send on it.

Mine must be broken, because mine CLEARLY does. Compare with a better interface or an analog mixer, the queue mix has a VERY noticeable delay.

[taturana]

funny... i have a motu ultralite here and it does what it says (even though though zero latency is definitely mktg exxageration)... the tiny conversion delay is not a problem,and no one who recorded here has ever complained about it... and i do work with some very touchy artists...

[theblue1]

Originally Posted by Gordon

Mine must be broken, because mine CLEARLY does. Compare with a better interface or an analog mixer, the queue mix has a VERY noticeable delay.

Using the 'direct' CueMix DSP, you should be able to get pretty close, probably a few ms -- depending on HW buffer settings, of course -- and that's a lot better than a roundtrip via FW through the computer. An analog board will give 'true' zero latency (nano or maybe microseconds as opposed to milliseconds) but no interface that uses a built in digital mixer (and, hence, the delays unavoidable in the AD/DA process) is going to get too far under around 2 ms.


OK... I've decided... I'm gonna run a latency test on the CueMix DSP later today I think. I've labored in darkness too long, guessing at a CueMix DSP latency of around 2 ms without actually testing... more later.

[Gordon]

I returned the MOTU piece of garbage for a Profire 2626 and couldn't be happier. maybe mine had something wrong with it but I just didn't feel like taking a chance.